Estimating Reliability After Corrective Action

A model is defined wherein corrective action may be accounted for in improving the estimation of reliability over the usual nominal success ratio. Probabilities for correcting any one of K failure modes which may arise are assumed known and a multinomial sampling procedure is discussed. Mean reliability is defined as a function of the unknown probabilities attached to the failure modes and the problem of estimating this parameter is posed. Two procedures which have been typically used in the past are discussed and some obvious weaknesses are pointed out. We then proceed to investigate several new estimators which we propose for the problem. No attempt is made to optimize with regard to a choice of estimators but properties of each of them are discussed which point to their limitations as well as usefulness in a given problem. Finally, we discuss some of the implications of the model we have assumed and propose other models that might be worthy of merit. Recommendations are made for further research in this timely and important area of investigation.

• 저널Management Science · 10(4) · 786–795
• 토픽Advanced Statistical Methods and Models · Statistics and Probability
• DOI10.1287/mnsc.10.4.786
• 저자William J. Corcoran, Harry Weingarten, Peter W. Zehna